Delayed & buggy integrated graphics drivers dog Intel and AMD

Delayed & buggy integrated graphics drivers dog Intel and AMD

Summary: More than a year after the launch of the Intel G965 integrated-graphics chipset and closing in on the launch of the G35 replacement chipset, Intel has finally released production non-BETA Vertex Shader enabled drivers that support half-way decent medium-resolution gaming.

SHARE:
TOPICS: Intel, Processors
80

More than a year after the launch of the Intel G965 integrated-graphics chipset and closing in on the launch of the G35 replacement chipset, Intel has finally released production non-BETA Vertex Shader enabled drivers that support half-way decent medium-resolution gaming.  These new drivers are only for Windows XP but Vista drivers won't launch until the end of this month which should be any day now, but should it really take 7 months after Vista launches for a product that's more than a year old?  Intel argues that this is essentially a $3 GPU on top of the cost of the motherboard and looking at the prices of G965 motherboards versus non-graphics P965 motherboards I'd say $3 is about right.  However, low cost shouldn't ever justify more than a year delay on 3D acceleration optimized drivers.

It isn't all bad news for Intel 965 drivers as the video playback was almost flawless on the HQV benchmark.  But as good as G965 DVD playback and HQV scores are, there are still screw ups on some DVDs with interlacing (especially anime or hand-drawn cartoon content for some reason) and I've been sending Intel screenshots.  When I tried to get official comments for this blog last week, Intel's initial response to me was that they're flawless on HQV and that my experience is isolated and not a driver problem.  I didn't care much for that response and I explained that this is a problem for an entire class of videos and they need to treat it as a driver issue because I can supply plenty of screenshots on my blog to prove my point from a wide variety of titles.

Intel eased on their position to "we're still looking at it" so I'm going hang on to this issue like a pitbull until they fix it.  I don't care if HQV says they aced the test, they failed my real world tests on some real world DVDs.  I will note that the current beta drivers have improved and the interlacing problem is less frequent than the production drivers so it's clear they can fix the problem; they just need to fix it ALL the time.

Note: It's not just Intel, the industry as a whole can't seem to get the de-interlacing problem fixed and this has been a problem since the late 90s on certain DVDs.  NVIDIA doesn't get a free ride either and they've been giving me plenty of interlacing problems on their lower-end video cards.  There's so much computational power in today's computers that I can afford to have my CPU utilization go from 6% to 12% if they'll just get the darn image quality right even if they have to run a line-doubler algorithm.

Another problem for Intel's video playback is captured HDV (1920x1080 1080i) content that I captured from my consumer-grade Sony HDR-HC1 camcorder.  While it's significantly better than an NVIDIA 6600 discrete graphics card in HDV playback, it's still jerky sometimes.  Intel says they can't replicate the problem yet so I'll have to put on my QA engineer's hat and give them an actual sample from my camcorder.  As HDV camcorders become prolific, this will become a bigger issue for consumers if they want to import their videos in to their media centers.  Since the CPU isn't even close to being maxed out, I don't understand why it's so hard to deliver some smooth HDV video playback.

On a related note, the BIOS update utility utterly failed to upgrade the Intel DG965WH motherboard.  I tested both the Windows utility and the Linux-based boot ISO with a burned CD and both failed.  The Windows utility rebooted the computer but gave an error message when it tried to update the BIOS so I tried the ISO method.  Unfortunately, the version of Linux Intel used doesn't support SATA based optical drives so I had to replace the drive with a PATA optical drive.  Once I booted up and ran the upgrade utility, same error message as before.  I've asked Intel for a fix but no luck there.

Intel needs to get their vertex shader drivers for Vista done as soon as possible and they need to address some of the remaining interlacing issues with standard DVD and stuttering issues with HD content.  Once the new G35 integrated graphics chipset comes out, the DX9 drivers should be complete with full Vertex Shader support but the wait for the DX10 (DirectX 10) drivers begins and we may have to wait several months after the launch of the product.  Intel would not comment on when DX10 drivers would be available, so the waiting game will soon start all over again.

<Next page - AMD 690G struggles with DVD playback quality>

AMD 690G struggles with DVD playback quality

AMD has their share of driver problems with the 690G integrated graphics motherboard chipset drivers.  While they have their 3D drivers working relatively well for an embedded graphics solution (though 10-25 fps in most situations isn't really playable) and working faster than Intel's beta Vista drivers with Vertex Shader support, their DVD video playback quality is horrible on the HQV benchmarks while Intel's G965 did very well on the HQV test.

Ironically, AMD actually claims on their website that Intel G965 fails the HQV Color bar test while ATI passes it when it fact it's just the opposite.  Here's the screenshot below of an AMD promotional video and I frankly don't know how they can get away posting it since AMD/ATI has the problem and not Intel on HQV.

I was starting to get tired of being a beta tester after I was given two ATI X1250 beta drivers for the 690G that promised to fix the HQV DVD playback problems but neither did the job completely.  The first beta driver might have fixed the 3:2 pull-down film conversion but nothing else was fixed.  AMD sent me a second set of private beta drivers that seem to have fixed the color bar resolution test and the cadence tests which is good but I could not get AMD to tell me when these fixes were going to be rolled in to production Catalyst drivers.  What I do know is that they shouldn't be pointing the finger at Intel when the G965 production drivers have been flawless on HQV for some time.

What didn't get fixed is the "jaggies", an artifact from interlaced video.  I've been complaining about this since last month when I posted screenshots and it is a huge problem with DVD playback.  Here are the screenshots again below.

Diagonal filter “jaggies” test:

When I tried to get comments from AMD last week, they spent more time talking about Intel's problems than addressing their own.  I had been asking about these HQV problems for over a month and it seemed like I had to explain the same thing to one of AMD's PR people all over again that there were DVD playback quality issues as if this was a surprise that there is even an issue.  I heard about how other people were getting great HQV results but I have a hard time believing that with the number of Catalyst drivers and beta drivers I've gone through.  I've even sent plenty of screenshots from the beta drivers to help ATI troubleshoot the issue and all I can hear is how wonderful and how much pride AMD takes in their drivers.  Finally I managed to get an official response and it read:

AMD 690 series platform technology provides an outstanding user experience for video playback in both standard-definition and high-definition formats. The interlacing issue that you report is limited to the HQV benchmark assessment tool and doesn’t impact the end-user experience for video playback. We are looking into a fix for this benchmark in a future release of our Catalyst software package. Rather than heavily focusing on benchmark optimizations in our drivers, our focus continues to be delivering great real-world user experiences with our technology. The AMD 690 is the only chipset to offer monthly WHQL-certified driver updates to maximize performance and compatibility on an ongoing basis.

Doesn't impact user experience?  I can't agree with that assessment at all with the number of real-world DVDs that I've tried.  While I am happy that AMD/ATI is doing something about this and they seem to be working hard to fix the problem, I would not be too happy with their response not just as a product reviewer but an AMD/ATI customer.  I do think the 690G platform has a lot of potential given low-cost of the product and the feature-set it offers but it would be nice if we can get some working drivers that don't have image quality issues.

690G shows potential with HD video playback: AMD's 690G drivers aren't all bad since it seems to handle HD playback (recorded from consumer HDV 1080i camera and imported via Firewire) fairly well but it completely botch standard DVD playback which doesn't make any sense to me.  If you can handle smooth HD playback, you should certainly have no problems with regular DVDs.  Intel on the other hand has a hard time handling HD playback.

AMD is also indicating that they can handle HD DVD and Blu-ray playback on their embedded 690G graphics chipset and that certainly sounds very exciting in a sub-$80 motherboard with HDMI output.  I'm in the process of testing that right now and I'll follow up with some Blu-ray playback results on 690G but I do wish AMD would fix their standard DVD playback issues since that's a more common usage scenario.

Conclusion: I realize we're talking about razor thin components here for both AMD and Intel, but I think it's only fair for consumers to expect completed drivers at the time of launch with all the promised features working.  The integrated graphics market dominates the overall graphics market and I don't think all those consumers who spent their hard-earned money are in the mood to listen to excuses.  The margins may be extremely thin but the volume is extremely high and plenty of money is made from the chipset market not to mention the fact that the chipset you choose determines whether you buy your CPU from Intel or AMD.

When either AMD or Intel shows progress, I'll be happy to heap praise upon them.  If neither delivers the fix, it's my job to make my readers aware of the issues and to shame the vendors publicly.  Consumers spend good money buying a computer and a fair chunk of that goes to Intel or AMD and they deserve to have good drivers.

[poll id=33]

<Return to top>

Topics: Intel, Processors

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

80 comments
Log in or register to join the discussion
  • Could you fix the 'jaggies' images test though?

    The images are of quite different widths, it's not really showing a like-for-like comparison.

    That said, it's good to see this highlighted. Intel dominates the graphics chip market with their integrated chipsets and centrino laptops, nVidia and ATI share only 46% of the total market between them, which makes casual games programming a PITA. Unless you want to shut out 35% of the market it's back to pixel-blitting on backbuffers and blitting the buffer to the display surface. Might as well be programming for DOS.

    In the meantime, this makes an essential part of Intel's lineup as they'd lose a lot of motherboard sales if they didn't provide these integrated chips on motherboards, people looking for cheap solutions (especially for business desktops) would go elsewhere. That $3 part can make a lot of $50 differences.
    odubtaig
    • The three bars are identical and that's all you need.

      No they're not different, the three bars are identical and that's the only part you need to see for this test. I only captured a larger portion of black space so I can type in some notes.
      georgeou
    • Oh I see your objection, here are some more screenshots

      Oh I see your objection, the AMD results are shown in 16x9 aspect ratio while the Intel screenshot is shown in 4x4 aspect ratio.

      If you want to see what the AMD screenshot looks like from beta 8.40.RC2 drivers in 4x3 aspect ratio, here's the screenshot. It doesn't look any better.

      http://blogs.zdnet.com/Ou/images/amd-jaggies-690g-840rc2.png
      georgeou
      • Typo correction

        Incorrect:
        Oh I see your objection, the AMD results are shown in 16x9 aspect ratio while the Intel screenshot is shown in 4x4 aspect ratio.

        Should have been:
        Oh I see your objection, the AMD results are shown in 16x9 aspect ratio while the Intel screenshot is shown in 4x3 aspect ratio.
        georgeou
        • Thanks for that fix

          I think I will start congratulating all ZDnet bloggers who fix their typographical mistakes. Perhaps that will inspire others.
          nucrash
    • "pixel-blitting on backbuffers"

      What are you on about? The programming is exactly the same for Intel as for NVIDIA/ATI.
      jinko
      • In theory only.

        If the graphics drivers could be relied on to handle everything properly this would be true, but they can't. You won't, For example, find many casual games sites keen to resell your game if it has any reliance on OpenGL (which is extremely useful for accelerating 2D graphics display) because there's not enough compatibility with the kind of shonky drivers the target customer base often has and it won't sell as well.

        As the comparison images show, you can't even rely on the integrated chips to use a decent anti-aliasing algorithm because it's cheaper and easier to develop a chip that only uses unweighted area sampling. So with a mind on selling to a market which has cheap computers, isn't interested in running Doom3 or Half-Life2 and still expects a basic 2D game to both look good and run smoothly you rely on software rendering and only use the hardware for basic display.

        In short: it's fast enough on any computer made in the last five years and it makes it [u]much[/u] more likely that your game will run on any PC out there without a hitch.
        odubtaig
  • Another typo.

    AMD has their share of driver problems with the 690G integrated graphics motherboard chipset drivers either.

    Glass houses...
    bjbrock
    • Thanks, fixed.

      nt
      georgeou
  • What is up with the poll

    I see 89% yes and 14% no. Do we have people vote that are from another dimension or some other plane that exceed the normal 100% that goes into consideration for the total number of voters?
    nucrash
    • Should have gotten the screen shot

      The problem is fixed. "YES" just doesn't update as fast as "NO"
      nucrash
    • LoL at the PoLL

      What's the point of the poll anyway? It seems rhetorical.
      xxn1927
      • duh factor

        of course they should "improve" the drivers. that's just normal ongoing QA
        richvball44
  • Where is the Nvidia review?

    All I see at my Best Buy is Nvidia graphics in HP and Gateway PCs. Can you do an assessment of those?
    truthteller33
    • Yes - I'm wondering about Nvidia too

      I've noticed that a number of lower end PCs using Athlon X2s have some variant of Nvidia GeForce 6100 for video. I wonder how it stacks up against Intel and AMD offerings.

      Perhaps George can update this entry in the future to include this popular video chipset.
      WiredGuy
      • Talk about late!

        Nvidia seemed to be the last to release Vista drivers. Heck, I can't even get Nvidia Vista SLI drivers yet for my 7800GTX cards in my SLI notebook! Nine month later. Someone told me about some hack where you spoof it into thinking it's a desktop... I give up. Grrrr LOL
        truthteller33
    • Will get to that, but even the NVIDIA 6200 and 6600 are bad

      Will get to that, but even the NVIDIA 6200 and 6600 dedicated cards are bad on video playback, even worse than the ATI HQV results and even worse on HD playback.
      georgeou
  • Intel can't even run Sims 2 kid's game

    This is real funny.... from a real journalist, too, from Businessweek.

    "A pretty river flows past Pleasantview. Its rippling waters reflect clouds and the graceful arches of a bridge. Unless, that is, you're playing Electronic Arts' (ERTS) The Sims 2 video game on an Intel (INTC) computer with underpowered graphics. If so, the water appears as a featureless patch of monochromatic blue, and many other graphic subtleties of the game are lost."

    http://www.businessweek.com/technology/content/aug2007/tc2007081_108723.htm?chan=search
    truthteller33
    • Not Intel's problem...

      This is a problem with the game, not the graphics chip.

      I make my living writing games a 3D graphics programs ans I can tell you that Intel graphics work fine (if a little slow). If you see weird graphics then it's most likely a problem with lazy game programmers, not Intels's chips/drivers.

      DirectX makes it easy to write games which don't run everywhere. There's pages and pages of "capability bits" which you have to constantly check and the documentation is terrible. Doing it right takes hard work, doing it wrong takes no effort at all.
      jinko
      • It certainly is Intel's problem

        Sims games are the top selling games right now. One could argue that the franchise is the most successful for everyday gamers ever in recent history.

        Why is is that only Intel has these problems with Sims, not ATI or Nvidia? Technological capability? No. Laziness? No. Arrogance? Yes. Even Intel's own blog said that the G965 wasn't made for gaming.

        In the end, it's the end user who gets the shaft. They bought their $1500 Centrino notebook at Best Buy with 965 graphics and it can't even play the Sims properly.
        truthteller33